AI & Content · September 22, 2026 · 8 min read
AI Content Tool Pricing Explained: Credits, Words, Seats, and Sites
Four billing meters, one comparable number: how to price AI content tools on cost per published post instead of the allowance on the sales page.
By FluxWriter Team
AI content tool pricing is built to resist comparison, and that is a design decision rather than an accident. Owners read the monthly number as the price, when the meter underneath it — credits, words, seats or connected sites — is what decides the bill. This breaks down the four billing models in the category, the contract terms that quietly move the total, and how to convert any plan into one comparable number.
Why the Advertised Price Is Not the Price
The number on the pricing page is an entry ticket. What you pay in month three depends on which unit the vendor chose to sell, and every unit in this category is picked to look generous at the moment of purchase.
Per-word allowances are the clearest case. A plan advertising 100,000 words a month reads as enormous. At 1,400 words a post that is 71 posts — more than most single-site owners publish in a year. The word ceiling was never going to be your constraint, so comparing plans on it tells you nothing.
The things that do run out sit elsewhere on the page. One connected site. Two seats. Images billed as a separate line. A research or competitor-analysis pass metered on its own, and rationed hard on anything under roughly $100 a month.
The advertised meter is always the abundant one. The scarce meter is printed in grey under the feature table. Find the scarce one first. Everything else on the page is decoration until you have.
The Four Meters Vendors Bill On
Almost every tool in this category bills on one of four meters, and the more expensive plans usually stack two or three of them together. Which unit you are buying matters more than the headline rate.
Score any plan you are considering across the four:
| Meter | What it really limits | Where it bites |
|---|---|---|
| Per word | Draft length, not draft count | Regenerating a section spends the allowance twice |
| Per credit | Everything, at a rate you cannot see | Images and research runs cost several credits each |
| Per seat | Headcount, not output | A reviewer logging in twice a month pays full price |
| Per site | Portfolio size, not volume | Site four often forces a tier built for 20 |
Stacking is the detail to watch. A plan metering credits and seats and sites is charging three times against the same output, and that structure scales worst exactly when your publishing volume starts to justify the spend.
Flat per-post pricing is the rare fifth pattern, because it moves the risk of an expensive draft onto the vendor rather than you. If one reaches your shortlist, price it first and make it the baseline every other plan has to beat.
What a Credit Actually Buys
A credit is a unit the vendor defines and can redefine at renewal. That is the whole problem with credit pricing — the unit floats while the price stays fixed.
In practice a credit maps to one operation, not one finished post. Illustrative arithmetic, not a quote from any vendor: a plain 1,400-word draft might spend 1 credit, while the same post with a research pass, a rewritten introduction, four generated images and one regenerated section spends 6 to 10. The ratio is the point. Real posts cost several times what the sample post in the demo cost.
Two habits protect you here. Run any trial on your worst case rather than your best — the long, research-heavy, image-heavy post you would genuinely publish. Then count what that single post consumed and divide the monthly allowance by it.
The answer is usually uncomfortable. Headline post counts assume the cheapest possible draft, so a plan advertised in the hundreds tends to deliver closer to half that at the settings you would really publish — and the gap is not a bug in the marketing. It is the marketing.
One more term hides in credit plans. Unused credits typically expire at the end of the billing month, so a quiet December does not fund a busy January.
Seats and Sites: The Quiet Multipliers
Seat pricing arrived from general business software, where every employee touches the tool daily. Content work does not look like that. You have one or two people producing and three or four who approve, comment, or read a draft once a fortnight.
Per-seat plans in this category typically run $20 to $60 a month per person, so four occasional reviewers can add $960 a year for maybe 30 minutes of use each month. Fix: before comparing plans, ask whether approve-only or read-only access is free. Some vendors give it away. Others charge full rate for a login that never generates a word.
Site limits punish a different buyer. Entry tiers usually connect exactly one site. The step up to three or five is normally the largest single price jump on the page, and extra sites past that are sold on top at $10 to $30 each.
That structure hurts portfolio owners most. Six niche sites publishing four posts each is 24 posts a month — less output than one busy company blog — but the site count alone pushes you into a tier priced for an agency. Count your properties first, then your posts. The site limit decides the plan more often than volume does, and it is the one ceiling you cannot publish your way around.
The Only Number Worth Comparing
Cost per published post is the number, and almost nobody calculates it. Take the monthly plan fee, add overage and any image or research add-ons, then divide by the posts you actually published — not the posts the plan allowed.
Utilisation is the biggest lever and the least discussed. Buyers pick a plan sized for the cadence they intend to hit, then publish a third of it, and the per-draft cost triples quietly while the invoice never moves.
Add the part no pricing page includes. Review, internal links and publishing take 30 to 60 minutes per post, which at a $50 hourly rate puts $25 to $50 of your own time on every piece — routinely more than the software line itself.
The meter decides how stable that figure is. A per-post plan holds $3.30 whatever the draft turns into. A credit plan does not, because the same $99 buys thirty plain drafts or about five research-and-image posts, so your cost per post rises every time you raise your own editorial standard.
So run the number on last quarter's real output, not the cadence you intend to hit. A $29 plan used twice a month costs $14.50 a post. A $99 plan used thirty times costs $3.30. The cheaper sticker is the more expensive tool, and no comparison table will ever show you that.
Terms That Move the Real Total
The contract terms move the annual number more than the tier choice does, and they are readable in about ten minutes.
Annual billing is usually sold as one or two months free. Take it only after two months on monthly, because fit in this category is close to impossible to judge from a demo. Lock in during week one and a $99 mistake becomes a $990 one.
Overage is priced at roughly 1.5x to 3x the in-plan rate, or it hard-stops you mid-month. Ask which. A hard stop on the 18th is worse than a surcharge if you have a publishing schedule to keep.
Three more clauses are worth finding before you buy. The first is whether a downgrade takes effect immediately or waits for renewal — waiting costs you a full month at the wrong tier. Refund windows run 7 days, 14 days, or nowhere at all. Then there is premium model access, which usually carries a multiplier on the standard credit rate and almost never appears on the comparison table.
Check the cancellation path too. Self-serve in the billing settings is the standard. Cancellation by email ticket predicts a retention conversation you did not want.
FAQ
How long a trial do I need to judge a plan?
Three real posts, not fourteen days. One draft tells you what a good day costs, and three tell you whether that number survives a topic that fights back. Most buyers spend the trial on easy sample subjects and learn nothing at all about the meter that will bill them.
How many words a month do I actually need?
Fewer than you think. Twelve posts a month at 1,400 words is under 17,000 words, which sits inside almost every entry plan sold. Buy for the constraint that binds instead — connected sites, seats, image generation, or the research runs your process depends on.
Should I pay per post or per month?
Per post is cleaner when volume is under 8 a month or genuinely unpredictable. Monthly subscriptions win above that, provided you use more than half the allowance you paid for. Below that line you are subsidising a cadence you are not keeping, and a smaller tier saves real money.
The Practical Takeaway
Price the constraint, not the headline. Write down the posts you published last month, the sites you need connected, and the number of people who need a login, then find the meter each shortlisted plan uses and the point where it runs out. Run the trial on your hardest post, count what it consumed, and divide the allowance by that. Convert everything to cost per published post with your review time included at your real hourly rate. Stay monthly for 60 days before signing anything annual. Start with the two plans you already have open in a tab.
If you are running that comparison now, tools like FluxWriter can help by metering in whole published posts rather than credits, and the free tier gives you 5 posts in total with no card, so you can price a real draft before you commit — but no billing model tells you whether 12 posts a month is the right cadence for your site.